The mainstream adoption of generative AI systems has raised the stakes around the issue of unverifiable quality. For lawyers and most clients the hard part with AI is verification, not performance. Generative AI is world changing but it is not currently possible to reliably predict when its mistakes will be trivial or obvious and when its mistakes will be catastrophic or not apparent. For investors the related problem is that verification of quality (in other words, what is investible) requires a mix of expertise and experience that is rare.